Spatio-temporal linear expansions for repolarization analysis

We propose a multichannel signal model based on linear expansions to analyze cardiac repolarization. Our hypothesis is that a joint spatiotemporal signal description which takes into account both temporal and spatial features provides a more compact signal representation, i.e. the signal energy is packed into a smaller number of coefficients. We also deal with the problem of estimating optimal basis functions in two different situations: when no information of the noise source is available and when the noise statistics is known or estimated from specific signals or intervals, like the TP segment.